Search Results (8,916)

Background/Objectives: Primary thyroid lymphoma accounts for approximately 0.2–2% of thyroid malignancies. Plasmablastic lymphoma (PBL), an aggressive large B-cell neoplasm with plasma-cell differentiation and frequent loss of conventional B-cell markers, is a rare thyroid presentation. We report a thyroid PBL presenting as an acute surgical airway emergency in an immunocompetent patient and highlight the diagnostic and management pitfalls that distinguish this from anaplastic thyroid carcinoma. Case Presentation: A 75-year-old man without any significant past medical history presented with rapidly progressive right-sided neck swelling, dysphagia, inspiratory stridor, and respiratory failure. Imaging demonstrated a large, thyroid-centered mass with tracheal involvement, initially raising concern for anaplastic thyroid carcinoma. Histopathology revealed a diffuse infiltrate of large, atypical cells with immunoblastic and plasmablastic morphology. The neoplastic cells were CD20- and CD138-negative but strongly MUM1-positive, with lambda light-chain restriction, bright CD38 by flow cytometry, a Ki-67 proliferation index exceeding 95%, aberrant cytoplasmic CD3 expression, and a MYC::IGH rearrangement, supporting a diagnosis of PBL. Staging identified extranodal perinephric disease and mesenteric lymphadenopathy, consistent with disseminated extranodal Ann Arbor stage IV disease. The patient underwent systemic treatment and initially had an excellent response; however, one month after the last treatment cycle they presented to the hospital with a mass consistent with recurrence. Discussion: Rapid growth, fixation, and tracheal invasion strongly suggest anaplastic thyroid carcinoma in routine clinical practice. However, plasmablastic lymphomas can present similarly and require fundamentally different treatment. In this case, loss of conventional B-cell markers, CD138 negativity, and aberrant cytoplasmic CD3 expression created substantial diagnostic challenges. Light-chain restriction, plasma-cell-associated markers, flow cytometry, and MYC cytogenetics were vitally important. Conclusions: Thyroid-presenting PBL is exceptionally rare and may closely mimic anaplastic thyroid carcinoma, including presentation with life-threatening airway compromise and tracheal involvement. This case highlights several diagnostic pitfalls: CD138 negativity despite plasma-cell differentiation, and aberrant cytoplasmic CD3 expression. Prompt airway stabilization, adequate tissue acquisition, broad immunophenotyping, light-chain assessment, flow cytometry, EBV/HHV8/ALK testing, and MYC cytogenetics are essential for accurate diagnosis and lymphoma-directed treatment. Full article

Background/Objectives: Interventional endoscopic ultrasound (I-EUS) has evolved from a diagnostic imaging modality into a transformative therapeutic platform encompassing biliary drainage, pancreatic interventions, luminal bypass, pain management, and ablative therapies. Despite the rapid proliferation of I-EUS research, a comprehensive bibliometric analysis characterizing the global intellectual architecture of this field is lacking. Methods: A Web of Science Core Collection search identified 4340 records, of which 2237 were included (2001–2024), analyzed with R bibliometrix (version 5.0) and VOSviewer (version 1.6.20). Results: Publication output demonstrated three distinct phases with pronounced acceleration from 2017. The United States led global output (n = 625, 27.9%), followed by Japan (n = 435, 19.4%) and Italy (n = 166, 7.42%). At the continental level, Asia collectively produced the largest share of output (n = 884, 39.5% of the total corpus), exceeding the Americas (n = 687, 30.7%) and Europe (n = 492, 22.0%). Gastrointestinal Endoscopy was the most productive journal (n = 173). Tokyo Medical University was the most productive institution (n = 93, 4.16%). Four thematic clusters were identified: EUS-guided biliary and pancreatic ductal drainage; diagnostic, ablative, and injection EUS for pancreatic tumors; LAMS-enabled luminal bypass and gallbladder drainage; and pancreatic fluid collections and necrotizing pancreatitis. Temporal overlay confirmed EUS-guided gastroenterostomy, EUS-guided gallbladder drainage, and EUS-guided radiofrequency ablation as the leading research frontiers. Annual output showed no discernible contraction during the COVID-19 pandemic period (2020–2021). Conclusions: The United States led global I-EUS research output with higher rates of international collaboration compared with East Asian nations, although Asia as a continent generated the largest aggregate volume of publications. EUS-guided biliary drainage and pancreatic fluid collection management constitute the established, high-volume core of the field, while EUS-guided gastroenterostomy and novel ablative technologies represent the most dynamic investigative frontiers. Full article

Medical image resolution plays a crucial role in early disease detection and fine-structure observation. Super-resolution reconstruction technology can restore low-resolution images to high-resolution versions, thereby assisting physicians in making accurate diagnoses. To address challenges in medical image super-resolution reconstruction, including insufficient global information acquisition, excessive network complexity, and suboptimal loss function adaptation for medical imaging data, this paper proposes an image super-resolution reconstruction algorithm named IDCASR-MMF based on improved dilated convolution and multi-module fusion. First, multi-dilation-rate dilated convolution is introduced to expand the receptive field and integrated with a spatial attention mechanism to dynamically calibrate high-frequency features after feature extraction. Subsequently, the Squeeze-and-Excitation module is fused with dilated convolution as a channel attention mechanism to streamline the network architecture. Finally, a weighted fusion strategy combining adversarial loss and MSE loss is adopted, where the dynamic adjustment of weighting coefficients balances pixel-level structural accuracy and high-frequency detail authenticity, achieving synergistic optimization of objective precision and subjective quality for medical images. To validate the effectiveness of the proposed algorithm, IDCASR-MMF is compared with 11 state-of-the-art methods across five datasets (Set5, Set14, BSD100, Urban100, and Bone FD). Experimental results demonstrate that the proposed algorithm achieves superior PSNR and SSIM values on multiple datasets, confirming that IDCASR-MMF can effectively reconstruct high-resolution medical images from low-resolution inputs. Full article

Semantic segmentation is a critical task in computer vision aiming to identify and classify individual pixels in an image, with numerous applications, for example, in autonomous driving and medical image analysis. However, semantic segmentation can be highly challenging, particularly due to the need for large amounts of annotated data. Annotating images is a time-consuming and costly process, often requiring expert knowledge and significant effort; moreover, saving the annotated images could dramatically increase the storage space. In this paper, we propose a novel approach for semantic segmentation, requiring only rough information about the proportions of individual semantic classes, hereafter referred to as semantic proportions (SPs), rather than the necessity of ground-truth segmentation maps. This greatly simplifies the data annotation process and thus will significantly reduce the annotation time, cost and storage space, opening up new possibilities for semantic segmentation tasks where obtaining the full ground-truth segmentation maps may not be feasible or practical. Our proposed method of utilising semantic proportions can (i) further be utilised as a booster in the presence of ground-truth segmentation maps to gain performance without extra data and model complexity, and (ii) also be seen as a parameter-free plug-and-play module, which can be attached to existing deep neural networks designed for semantic segmentation. Extensive experimental results demonstrate the good performance of our method compared to benchmark methods that rely on ground-truth segmentation maps. Utilising semantic proportions suggested in this work offers a promising direction for future semantic segmentation research. Full article

Dental caries remains one of the most prevalent oral health problems worldwide, yet automated detection in intraoral photographs is challenging because of variable lighting, specular reflections, saliva, restoration margins, and subtle early demineralization. We propose WaveViT-YOLO, a hybrid architecture built on a YOLOv9m backbone and integrating (i) a discrete wavelet transform (DWT) preprocessing stage, (ii) learnable WaveletAttention modules at the feature-pyramid scales, and (iii) ViT-based MultiScaleCrossAttention fusion. On the publicly available Annotated Intraoral Image Dataset (6313images; patient-level 70/15/15 split; three independent seeds), YOLOv9m is the strongest standalone YOLO model by mAP@50 (mAP@50 = $0.807\pm 0.003$; mAP@50–95 = $0.642\pm 0.004$). WaveViT-YOLO achieves the highest measured mAP@50 ($0.814\pm 0.005$; $0.007$ absolute and $+0.87$% relative), mAP@50–95 ($0.647\pm 0.004$), F1 ($0.831\pm 0.005$), and PR-AUC ($0.845$) among the evaluated models. The model contains 26.3 M parameters, a $30.8$% increase over the 20.1 M YOLOv9m baseline. Model-only inference is $27.3\pm 1.6$ ms on an NVIDIA T4 GPU, while the current CPU DWT stage adds $235.2\pm 8.0$ ms, giving approximately 262.5 ms/image end-to-end; therefore, the current pipeline is not real-time end-to-end. Using the displayed seed-averaged mAP@50 values, the isolated relative changes are $+0.62$% for DWT and $+0.37$% for either WaveletAttention or ViT fusion, whereas the full configuration reaches $+0.87$%. The paired three-seed comparison against YOLOv9m yields $t\left(2\right)=6.06$, $p=0.026$, and Cohen’s $d_z=3.50$; because $n=3$, this analysis is treated as exploratory. Small lesions (<$0.098$% image area) remain the principal limitation (recall = $0.477$). Because evaluation uses clinician-provided annotations from one retrospective dataset and no independent external or prospective validation was completed, the system is presented as a research-stage screening architecture rather than a clinically validated diagnostic tool. Full article

Accurate morphological analysis of blood smears is vital for hematological diagnosis, yet manual examination is labor-intensive and subjective. While deep learning offers automation, its black-box nature and computational demands often hinder clinical trust and deployment. We propose XHIC-Net, an Explainable Hybrid Involution–Convolution Network designed for efficient and transparent cell classification. By integrating spatially adaptive involution operations with convolutional layers within a residual framework, XHIC-Net captures both contextual and fine-grained features efficiently. To enhance interpretability, a Grad-CAM-based explainable AI (XAI) module visualizes the cellular regions driving model predictions. The proposed framework was evaluated on a dataset comprising 12,879 microscopic blood smear images belonging to 12 morphological cell categories. Experimental results demonstrate that XHIC-Net achieves an overall accuracy of 98.88%, precision of 98.89%, recall of 98.87%, F1-score of 0.9887, and Cohen’s Kappa score of 0.9887. It outperformed established models, including DL models such as EfficientNetV2S, MobileNet family, DenseNet family, and VGG16, while using fewer parameters and requiring shorter training times. Furthermore, the XAI maps consistently highlighted biologically relevant structures, validating the model’s decision-making process. XHIC-Net is a strong, effective, and clear research model for automated hematology. With future clinical validation, it has the potential to be modified for point-of-care diagnostics in healthcare settings with limited resources. Full article

Background/Objectives: Forensic age estimation increasingly relies on non-ionizing magnetic resonance imaging (MRI) evaluation of the knee growth plates. The original five-stage classification proposed by Dedouit et al. was developed using spin-echo proton-density-weighted imaging, and its applicability to gradient-echo sequences with different contrast mechanisms has not been fully established. This study aimed to apply the original, unmodified Dedouit classification to a volumetric three-dimensional Multiple Echo Data Image Combination (3D MEDIC) sequence and to generate corresponding age thresholds in a Turkish sample. Methods: Knee MRI examinations of 309 individuals (153 males, 156 females; age range 9.18–25.97 years) were retrospectively evaluated at a 3-Tesla field strength. Two experienced observers independently staged the distal femoral and proximal tibial epiphyses; intra- and inter-observer agreement were assessed using Cohen’s kappa. Spearman’s correlation and the Mann–Whitney U test, with rank-biserial effect sizes and Bonferroni correction, were used to assess the relationship between age and stage and between-sex differences, respectively. To provide forensically applicable thresholds, we additionally modeled the age at which the probability of complete fusion (Stage V) reached 50%, with 95% bootstrap confidence intervals, and calculated the sensitivity, specificity, and area under the curve (AUC) of Stage V for identifying individuals ≥18 years. Results: Agreement was very good for both epiphyses (kappa = 0.808–0.833). Age correlated strongly with stage for both the femur (rho = 0.70–0.76) and tibia (rho = 0.68–0.74). The 50%-probability age for complete fusion ranged from 15.23 years (tibia, females) to 17.43 years (femur, males). Stage V showed high sensitivity (0.96–0.98) but markedly lower and sex-dependent specificity for the 18-year threshold (0.48–0.58 in females versus 0.77–0.85 in males), indicating that Stage V alone is a poor sole criterion for confirming adult status in females. The youngest age at which Stage V was observed was 14.67 years in females and 16.07–16.16 years in males, markedly younger than thresholds reported using spin-echo imaging in the original Dedouit cohort. However, as an extreme-value statistic based on a single individual, this minimum should not be used as a forensic threshold in isolation. Conclusions: The Dedouit classification remains reproducible when applied to a 3D gradient-echo sequence, but the resulting age thresholds may be influenced by acquisition technique, population-specific factors such as socioeconomic status, or both. Consequently, thresholds derived from different MRI sequences or populations should not be assumed to be interchangeable without local validation. Full article

Objective: Our objective was to describe morphologic features and longitudinal changes observed on high-resolution vessel wall imaging (HR-VWI) in a selected retrospective case series of patients evaluated for suspected inflammatory intracranial vasculopathy. Methods: This retrospective case series included patients who underwent 3T intracranial HR-VWI for suspected inflammatory intracranial vasculopathy. Clinical and imaging data were collected from medical records and imaging archives. Three radiologists reviewed baseline and follow-up examinations for vessel wall enhancement (VWE) pattern and grade, wall thickness, luminal narrowing, arterial distribution, diffusion-restricted lesions, and susceptibility-sensitive findings. Analyses were descriptive and patient-level. Each patient contributed one baseline and one final examination to longitudinal summaries; intermediate scans characterized individual trajectories only. Results: Nineteen patients were included, and 17 underwent serial HR-VWI. Median age was 46 years (IQR, 35.5–60.0 years), and 11 patients were female. Follow-up HR-VWI was available in 17 patients over a median of 391 days (IQR, 237–1819 days; range, 15–4034 days). VWE was present in all patients, with a purely concentric pattern in 15/19 (78.9%). The middle cerebral and internal carotid arteries were involved in 17/19 and 13/19 patients, respectively. Among the 17 patients with paired examinations, VWE grade decreased in 13 patients (76.5%) and remained stable in four patients (23.5%), with the median decreasing from 3.0 to 1.0. Maximum wall thickness decreased in 16 patients and remained stable in one, with the median changing from 2.0 mm to 1.4 mm. Conclusions: In this selected retrospective case series of patients evaluated for suspected inflammatory intracranial vasculopathy, 3T HR-VWI commonly demonstrated concentric VWE and provided descriptive longitudinal information regarding enhancement evolution. These findings should not be interpreted as evidence of diagnostic performance. Full article

AI-generated synthetic media are evolving from tools of digital manipulation into a practical antidote for persistent challenges in digital health implementation. However, the computational capabilities that characterise these technologies, their applications and downstream health-related benefits remain fragmented and insufficiently synthesised. This systematic review identified the computational capabilities that characterise AI-generated synthetic media in health, examined their applications and benefits, and developed an integrative framework linking these domains. Twenty-four studies published between 2021 and 31 May 2026 were included. Methodological quality was assessed using the Mixed Methods Appraisal Tool (MMAT) and findings were synthesised through thematic analysis. The synthesis revealed an integrated set of capabilities spanning photorealistic medical-image generation, modality-specific synthesis of clinical images and physiological signals, synthetic non-image health-data creation, preservation of statistical distributions, temporal patterns and clinical relationships, generation of diverse, novel and non-memorised samples and controlled transformation of medical and audiovisual content. Privacy-oriented synthesis and deepfake detection emerged as distinct components supporting privacy-conscious data use, clinical verification and healthcare safety. These demonstrated capabilities were linked to empirically evaluated and indicated applications, including data augmentation, AI model training, diagnostic model development, privacy-oriented health-data sharing, medical education, patient-facing communication, therapeutic support, clinical safety, health-system analytics and planning. The resulting Computational Capability–Application–Benefit (CAB) Framework conceptualises synthetic media as an evidence-graded pathway distinguishing demonstrated computational capabilities, evaluated health-related applications and reported, indicated or potential downstream benefits requiring further validation. AI-generated synthetic media, therefore, represent an emerging computational infrastructure with potential to support safer, privacy-conscious, adaptive and data-intensive healthcare. Full article

Background: Skin cancer encompasses a diverse range of malignancies and remains a significant global health challenge. Accurate machine-learning-assisted diagnosis can substantially improve patient outcomes through early detection and timely clinical intervention. Federated Learning (FL) enables privacy-preserving collaborative model training across multiple healthcare institutions while ensuring that sensitive patient data remain decentralized. However, deploying advanced architectures such as Vision Transformers (ViTs) in clinical environments is challenging due to the high computational demands of self-attention mechanisms. Methods: This work proposes FedSwin-LHTP, an efficient federated learning framework for skin lesion classification that integrates a Swin Transformer backbone with a Lightweight Hessian-Inspired Token Pruning (LHTP) mechanism. LHTP estimates token importance using a second-order Taylor approximation around converged local model parameters to identify less informative patch tokens, enabling the early pruning of redundant representations without explicitly computing the Hessian matrix. Furthermore, the framework incorporates the FedProx optimization objective to mitigate client drift under heterogeneous non-IID data distributions. The proposed framework is evaluated on the HAM10000 and ISIC datasets under realistic non-IID federated settings. Results: Experimental results demonstrate stable convergence, effective knowledge aggregation, and robust diagnostic discrimination across distributed clients. By adaptively pruning approximately 60% of Stage-1 tokens, the proposed framework substantially reduces the computational burden of local transformer processing while maintaining high multiclass classification performance, achieving an accuracy of up to 96.1% on the evaluated datasets. Conclusions: These results highlight the potential of FedSwin-LHTP as a practical, privacy-preserving, and resource-efficient solution for collaborative healthcare intelligence. Full article

Systolic anterior motion (SAM) of the mitral valve can occur either in association with or in the absence of hypertrophic obstructive cardiomyopathy (HOCM). SAM induces dynamic left ventricular outflow tract obstruction (LVOTO) and, in the majority of cases, is associated with a substantial degree of mitral regurgitation (MR) that significantly impacts patient morbidity and mortality. This narrative review explores the contemporary understanding of the pathophysiology, diagnosis, and management of SAM, focusing particularly on surgical strategies and the novel therapeutic class of cardiac myosin inhibitors. Extended septal myectomy remains the gold-standard treatment for HOCM-related SAM, yielding superior long-term outcomes compared to alcohol septal ablation (ASA). Advanced imaging modalities, including three-dimensional transesophageal echocardiography (3D-TEE), enable precise pre-operative characterization of the mitral valve apparatus. Some patients may benefit from septal reduction strategies while concomitant mitral valve interventions are generally reserved for highly selected cases with intrinsic valve pathology or persistent residual SAM, thereby avoiding unnecessary valvular manipulation and its potential hemodynamic risks. Mavacamten, a selective cardiac myosin inhibitor, represents an important advance in pharmacological management, achieving a mean LVOT gradient reduction of 37.2 mmHg in symptomatic patients. Furthermore, data from the MARVEL registry confirm the real-world clinical efficacy of mavacamten in obstructive hypertrophic cardiomyopathy, with 86% of patients successfully down-staged to NYHA functional class I–II. Although ASA serves as a viable alternative to surgery, it entails a higher risk of conduction abnormalities requiring permanent pacemaker implantation and subsequent re-intervention. Beyond classical hypertrophic SAM, this review addresses the diagnosis and management of post-mitral repair complications and non-hypertrophic variants. Optimal management and risk stratification remain an evolving field requiring a multidisciplinary Heart Team approach, leverage of advanced imaging, and adoption of novel medical therapies. Interventional strategies must be carefully tailored to maximize the efficacy-to-safety profile on an individualized patient basis. Full article

Cardiovascular ultrasound is a cornerstone of noninvasive cardiac and vascular assessment, yet conventional interpretation remains operator-dependent, variable, and limited in sensitivity for subclinical disease. Artificial intelligence (AI), particularly machine learning (ML), deep learning (DL), and, most recently, vision–language and foundation models, offers tools to automate, standardize, and extend ultrasound analysis. This narrative review examines the role of AI-enhanced cardiovascular ultrasound in the transition from descriptive imaging toward predictive and personalized medicine. We conducted a structured literature search of PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar (January 2016–May 2026), combining Medical Subject Headings and free-text terms related to AI and cardiovascular ultrasound. Original studies, meta-analyses, reviews, consensus documents, and seminal works were considered. AI now spans the entire echocardiographic workflow, acquisition guidance, view classification, segmentation (Dice ≈ 0.92–0.94 on public datasets), and automated quantification of ejection fraction and global longitudinal strain, achieving expert-level accuracy and improved reproducibility. Across clinical domains, AI supports ischemia detection on stress echocardiography, heart-failure phenogrouping, Doppler-independent aortic stenosis detection, and carotid plaque characterization for stroke-risk stratification. Emerging vision–language and multitask foundation models (e.g., EchoCLIP, EchoPrime, and PanEcho) point toward general-purpose interpretation, and a growing number of tools (Caption Guidance, Us2.ai, and Ultromics EchoGo) have obtained FDA clearance and/or CE marking. Increasingly, AI-derived imaging biomarkers feed multimodal models that enable individualized risk prediction and therapy selection. AI-enhanced cardiovascular ultrasound is poised to become a central tool of precision cardiology. Realizing its potential will require prospective multicenter validation, cross-vendor standardization, attention to generalizability, interpretability, and reproducibility, and evolving regulatory and ethical frameworks. Full article

Background and Clinical Significance: Parkinson’s disease (PD) is a progressive neurodegenerative disorder for which nonpharmacological neuromodulatory approaches are being explored as adjunctive strategies. Deep transcranial magnetic stimulation (dTMS) using an H-coil can engage broader and deeper cortical networks than conventional focal coils, although anatomically distant structures are expected to be influenced through network modulation rather than direct electromagnetic stimulation. Case Presentation: We describe a five-patient exploratory case series evaluating clinical measures and cerebral perfusion before and three months after high-frequency dTMS targeting the supplementary motor area (SMA). Clinical outcomes included the Unified Parkinson’s Disease Rating Scale (UPDRS) Parts I-IV, Non-Motor Symptoms Scale (NMSS), Hoehn–Yahr stage, and Timed Up and Go test. Antiparkinsonian medication regimens and doses remained unchanged during the three-month follow-up. Cerebral perfusion was assessed using single-photon emission computed tomography (SPECT) and statistical parametric mapping. No clinical outcome reached nominal statistical significance at follow-up; given the sample size, this should be interpreted as limited statistical power rather than evidence of no effect. At an exploratory voxel-wise threshold of p < 0.001, uncorrected, SPECT identified a 28-voxel cluster of increased regional cerebral blood flow in the right cerebellar cortex (peak MNI coordinates: 22, −38, −44; t = 5.96; z = 3.11). Conclusions: These observations are hypothesis-generating and may reflect network-level modulation of SMA–cerebellar circuitry. Because only five patients were included, no sham-controlled arm was available, and the imaging analysis used an uncorrected exploratory threshold, causal or efficacy claims cannot be made. Larger sham-controlled studies with standardized dose-to-imaging timing, formal neuropsychological assessment, individualized targeting, and prespecified corrected imaging analyses are warranted. Full article

Background: Elevated plasma homocysteine (HCY) is a risk factor for Alzheimer’s disease (AD), but its relationship with structural brain changes across the AD continuum remains unclear. The choroid plexus (CP) regulates cerebrospinal fluid homeostasis and may interface with peripheral metabolic signals. Whether HCY relates to CP structural alterations and disease progression remains unknown. Methods: We analyzed 819 Alzheimer’s Disease Neuroimaging Initiative (ADNI) participants (229 cognitively normal (CN), 397 with mild cognitive impairment (MCI), and 193 with AD dementia). Multinomial logistic regression assessed associations between HCY and diagnosis under stepwise covariate adjustment. Phenotype-wide structural magnetic resonance imaging (MRI) mapping identified HCY-associated signals. Cox models evaluated associations of CP volume (CPV) with CN-to-MCI and MCI-to-AD dementia conversion and whether CPV added prognostic discrimination beyond baseline disease-severity markers. Independent human CP single-nucleus and spatial transcriptomic datasets were reanalyzed to characterize epithelial expression states and their spatial organization in a hypothesis-generating analysis. Results: Higher HCY was associated with MCI and AD dementia; however, the AD association attenuated after adjustment for renal function, vitamin B12, and medications, whereas the MCI association remained stable. CPV was among the HCY-associated MRI signals that persisted after progressive covariate adjustment. Right and bilateral CPV showed model-dependent associations with MCI-to-AD dementia conversion. In the disease-severity sensitivity analysis, larger right and bilateral CPV remained associated with a higher risk of progression from MCI to AD dementia. Single-nucleus analysis identified two CP epithelial states with relatively high expression of one-carbon metabolism-related genes, termed one-carbon metabolism-enriched epithelial state A (OCM-Epi-A) and state B (OCM-Epi-B). Donor-level pseudobulk analysis did not identify pathway enrichment after false discovery rate correction, whereas OCM-Epi-A–like spots were located near endothelial spots more often than expected by chance in three of the four spatial samples. Conclusions: Circulating HCY was associated with larger CPV, and larger CPV showed model-dependent associations with MCI-to-AD dementia progression. Independent transcriptomic reanalysis identified one-carbon metabolism-enriched epithelial states and their spatial organization in postmortem CP tissue, providing hypothesis-generating tissue-level context for the ADNI associations. Full article

Background/Objectives: Paget’s disease of bone (PDB) and osteoporosis are chronic metabolic bone disorders characterized by disrupted bone remodeling and increased skeletal fragility; however, the underlying mechanism of PDB remains poorly understood. Artificial intelligence (AI) has emerged as a transformative tool in medical imaging, enabling automated feature extraction and improved diagnostic classification of skeletal disorders. This study aimed to investigate whether AI could distinguish subtle variations in bone morphology between PDB and osteoporotic bone. Methods: C57BL/6 mice femurs were scanned by µCT: 16 optineurin-knockout mice with a PDB phenotype (20–26 months), 25 genetically matched wild-type Aging mice (20–26 months), and 15 ovariectomized (OVX) mice with osteoporotic bone phenotype (4.5 months). Two AI algorithms were investigated: a machine learning (ML) model using 22 µCT-derived features trained with a Random Forest (RF) classifier, and a deep learning (DL) model using a 3D convolutional neural network (3D-CNN) trained on raw µCT images. Leave-one-out cross-validation was applied to evaluate model robustness. Results: Significant differences in volumetric, density, and morphological parameters of cortical and trabecular bone were observed between PDB and osteoporosis (p < 0.05). The RF algorithm achieved 90% accuracy in distinguishing PDB from both aging- and OVX-induced osteoporosis and provided feature importance rankings that improved model interpretability. The 3D-CNN achieved classification accuracies of 70% for PDB vs. OVX and 68% for PDB vs. aging, demonstrating the feasibility of an image-based DL approach. Conclusions: AI-based RF and 3D-CNN models demonstrated promising performance in differentiating PDB from osteoporosis using µCT-derived bone features. These findings suggest potential for using AI to assist with analyzing skeletal images in the diagnosis of metabolic bone disorders. Full article